So here's what's happening. For nifti, all 185 volumes for each run are
stored as one 4D image. The SPM file selector, by default, only selects the
first volume in a 4D image. So while you are trying to input 5 sets of 185
volumes, you are in fact only inputting 5 volumes.
There's an easy enough way to select all 185 volumes. To do it, change the
'1' under the filter box to '1:185' and then select all the resulting volumes.
That said, this still WILL NOT give you proper realignment results! In 4D
nifti files, all of the volumes share one common header. Because
realignment works by changing the volume headers *relative* to each other,
realignment will NOT work if the volumes do not have separate headers! For
4D files, realignment will look like it's executing perfectly, but at the
end all the volumes in any 4D file will be in the same orientation as the
last volume in the file. Perhaps reslicing fixes the problem (perhaps it
doesn't), but I never reslice during realignment anyways. Ultimately, in
order to perform a proper realignment, each 4D file must be decomposed into
separate single volume (3D) nifti files (185 of them in your case). I use
fslsplit to do this, but perhaps there is some way native to SPM to do it as
well. Once all the files are in 3D nifiti format, you can carry out
realignment just as you did using analyze format.
Hope this helps,
-Neil
Neil Chatterjee
Research Assistant
Stanford University Systems Neuroscience and Pain Lab
[log in to unmask]
(650)-724-0522
On Fri, 26 Dec 2008 08:22:05 -0500, Yune Lee <[log in to unmask]> wrote:
> Dear SPMers,
>
> I'm thinking of converting from analyze format to nifti format and as a
>first attempt, I compared the realignment results with nift format to that
>with analyze format.
> Turns out the realignment with nifti not only takes too quick (a couple of
>seconds) but also looks quite different from that with analyze format.
> The big difference is that there are only 5 nifti files (one for each run)
>wheras there are 925 analyze files (185 per each run) for the 5 runs of the
>same subject.
> So, SPM5 seems to think that there is only 5 files reardless of the file
>format. If the results look consisten, nonetheless, I would assume that it's
>all good. But,
> I'm not sure if the way I did was correct (I took the same step of
>realignment that I used to do with analyze format. : select all files->
>estimate & reslice-> leave others as default).
> Attached are the results of two different formats. I wonder how people do
>preprocessing using nifti format and any help would be much appreciated.
>
> Thanks,
> YSL
>
|